Greyscale, black and white or sepia, with a live preview. Runs on your device.
Sepia is the warm brown tone of old studio portraits and faded family prints. People add it to a modern photo for a vintage look, to make a scan feel older, or to match a set of pictures for an album or a poster. This page opens with the Sepia style selected, so the preview shows the tinted version as soon as a picture is added. The look comes from one fixed colour matrix, the same for every photo. There is no strength slider and the grey method setting has no effect here, so what you see in the preview is what you get. A dark picture comes out deep brown and a bright one comes out cream. Your original file is never touched, and the result is saved under a new name ending in -sepia, so keep the original if you may want the colours back. A JPEG is saved again at quality 92, and lossy encoding can change some detail. If a softer tint is what you want, the greyscale style is the closer starting point.
Greyscale replaces each pixel's red, green and blue with one grey level. The tool offers three ways to choose that level, and they disagree most on strong colours. Take one pure red pixel. Luminance 54: it weighs green most and red much less, the way the eye does, so red turns dark. Average 85: it adds red, green and blue equally and divides by three, so red lands at one third of full brightness. Lightness 128: it takes the midpoint of the brightest and the darkest channel, so red lands in the middle of the scale. Luminance is the default because it keeps skin, sky and foliage looking natural in a photo. Average makes greens darker and blues lighter. Lightness looks flat. The preview shows the difference on your own picture before you save anything.
Greyscale keeps every shade between black and white. Black and white keeps only two: each pixel becomes pure black or pure white. The tool first works out the grey level with the method you chose, then compares it with the threshold, a number from 0 to 255. A pixel at or above the threshold turns white and anything below turns black. The default is 128, the middle. A pure red pixel is black at 128 with Luminance, because its grey level is 54. Raise the threshold and more of the picture turns black. Lower it and more turns white. This gives the hard, high-contrast look of a photocopy or a stamp, and it suits logos and line drawings better than portraits. It does not dither, so smooth gradients become flat bands.
Sepia gives the warm brown tint of an old print. It mixes the red, green and blue of each pixel into a new red, green and blue with fixed weights, so every picture gets the same treatment and there is no strength slider. The grey method setting does not apply to sepia. A white pixel becomes a warm off-white, and a mid grey becomes a mid brown. If the result is too strong for you, this tool cannot soften it. Convert to greyscale instead, or keep the original beside it.
The picture is drawn fresh on a canvas in this tab, filtered in strips of 256 rows so a very large image never needs one huge block of memory, and saved again in the same format as the original. JPEG and WebP files are saved again at quality 92. Both use lossy encoding, so some detail may change. PNG is saved without loss. Transparent areas stay transparent: only the colour of each pixel changes, never its opacity. Because the picture is redrawn, location and camera details stored in the original are not carried over, and the original file on your device is never changed. The filter is plain whole-number arithmetic, so the same decoded pixels always give the same result. Browsers can decode the same file to slightly different pixels, so a saved file may still differ a little between browsers. This tool does not do colour-managed or gamma-linear conversion, dithering, curves or per-colour mixing, and it does not change the file format. If a picture is too large for this device's canvas it is made smaller, and the result shows both sizes.
No. The tint is one fixed colour matrix and there is no strength slider. A pure white pixel always becomes 255, 255, 239, a warm off-white. For a lighter effect, use the Greyscale style instead.
A little. The picture is drawn again and saved at quality 92, and lossy encoding can change some detail. A PNG is saved without loss, so choose a PNG original when you plan to edit the sepia picture further.
Not from the saved file, because the original colours are mixed into brown and cannot be recovered. The tool never changes your original, so keep it and run the filter again with other settings if you change your mind.
Luminance, which is the default. It uses the Rec. 709 weights of 0.2126 for red, 0.7152 for green and 0.0722 for blue, so green foliage stays bright and a blue sky goes dark, as the eye sees them. Average and Lightness are there for the cases where you want a different look, such as flat graphics.
Each method asks a different question. For pure red, Luminance gives 54 because red adds little to perceived brightness, Average gives 85 because it counts the three channels equally, and Lightness gives 128 because it is halfway between the brightest channel, 255, and the darkest, 0. Try the three on a red object in the preview.
It only matters in Black and white. Every pixel whose grey level is at or above the threshold becomes white, and every other pixel becomes black. At the default of 128, a mid-bright picture splits about evenly. Type 0 and everything is white, type 255 and only pure white stays white. Anything outside 0 to 255 is corrected to the nearest end when you leave the field.
Yes. A PNG or WebP with see-through areas stays see-through, because only the red, green and blue of each pixel are changed and its opacity is left alone. A JPEG has no transparency to keep.
The tool never overwrites your file, it gives you a new one named with -bw or -sepia, so the original stays where it was. Keep it if you may want the colours back. A black and white or sepia file cannot be turned back into colour.
No. The filter runs on a canvas inside this tab, so no server receives your picture. The page counts visits with Google Analytics using rough groups such as 1 to 10 MB, never a file name, an image, or the style you picked.
The format stays the same: a JPEG stays a JPEG, a PNG a PNG and a WebP a WebP. The pixel size stays the same too, unless the picture is bigger than this device's canvas allows. Then it is made smaller to fit, and the result shows the original size and the new one, in the form width × height → width × height px.